Journal of Biomechanics 48 (2015) 1950–1956
Contents lists available at ScienceDirect
Journal of Biomechanics journal homepage: www.elsevier.com/locate/jbiomech www.JBiomech.com
Effects of unweighting and speed on in-shoe regional loading during running on a lower body positive pressure treadmill James M. Smoliga a,n, Leah Anne Wirfel b, Danielle Paul b, Mary Doarnberger b, Kevin R. Ford a a b
Human Biomechanics and Physiology Laboratory, Department of Physical Therapy, High Point University, High Point, NC 27268, USA Department of Exercise Science, High Point University, High Point, NC 27409, USA
art ic l e i nf o
a b s t r a c t
Article history: Accepted 3 April 2015
The purpose of this study was to determine how unweighted running on a lower body positive pressure treadmill (LBPPT) modifies in-shoe regional loading. Ten experienced runners were fit with pressure distribution measurement insoles and ran at 100%, 120%, and 140% of self-reported easy training pace on a LBPPT at 20%, 40%, 60%, 80%, and 100% body weight percentage settings (BWSet). Speeds and BWSet were in random order. A linear mixed effect model (po0.05 significance level) was used to compare differences in whole foot and regional maximum in-shoe plantar force (FMAX), impulse, and relative load distribution across speeds and BWSet. There were significant main effects (po 0.001) for running speed and BWSet for whole foot Fmax and impulse. The model revealed 1.4% and 0.24% increases in whole foot FMAX (times body weight) and impulse, respectively, for every unit increase in body weight percentage. There was a significant main effect for BWSet on Fmax and relative load (p o0.05) for each of the nine foot regions examined, though four regions were not different between 80% and 100% BWSet. There was a significant (po 0.001) main effect for BWSet on forefoot to rear foot relative load. Linear relationships were found between increases in BWSet and increases in-shoe Fmax and impulse, resulting from regional changes in foot pressure which represent a shift towards forefoot loading, most evident o 80% BWSet. Estimating in-shoe regional loading parameters may be useful during rehabilitation and training to appropriately prescribe specific speed and body weight levels, without exceeding certain critical peak force levels while running. & 2015 Elsevier Ltd. All rights reserved.
Keywords: Running Unweighting Unloading Treadmill Overuse injury rehabilitation Stress fracture
1. Introduction Applied biomechanics experts regularly adopt new technology to optimize musculoskeletal rehabilitation and maximize athletic performance. While various methods to reduce musculoskeletal loading during ambulation are available, such as underwater treadmills and harness systems (Newman et al., 1994), lower body positive pressure treadmills (LBPPT) provide a novel method of unweighted exercise and are becoming more readily available in rehabilitation and sports performance settings. Reduced foot and overall loading on the lower extremity may be beneficial to those rehabilitating from certain overuse injuries (Samaan et al., 2014; Warden et al., 2014). For instance, a graded return to running program would be designed to not only reduce the maximum ground reaction force (GRF) but also the total force time-integral during each stance phase. However, n Correspondence to: Associate Professor of Physiology, Department of Physical Therapy 833 Montlieu Avenue, High Point University, High Point, NC 27268. Tel.: þ 1 336 841 9480; fax: þ1 336 888 6394. E-mail address: [email protected] (J.M. Smoliga).
http://dx.doi.org/10.1016/j.jbiomech.2015.04.009 0021-9290/& 2015 Elsevier Ltd. All rights reserved.
certain individuals that may benefit from reduced plantar loading during rehabilitation, such as those returning from a metatarsal stress fracture, may alter running mechanics and potentially not reduce explicit plantar regions. Flynn et al. (1997) identified a nonconsistent reduction of regional peak pressure during running using an unweighted harness system. However, the force and time integral were not analyzed regionally, nor were measurements across multiple running speeds or unweighting magnitudes reported. Therefore, to ensure appropriate use during rehabilitation and training, it is important to understand specific alterations in regional plantar loading under various conditions. When exercising in LBPPT, airflow is directed into a capsule surrounding a treadmill and lower extremities of the athlete or patient (Fig. 1). Air pressure is increased within the capsule which applies an upward directed vertical force on the user. Therefore, the net GRF (Eq. (1)) is reduced as the downward gravitational force (M g, mass gravity) by the upward directed force from the air pressure differential (ΔP) multiplied by the cross-sectional area of the lower body and seal (Abody) (Cutuk et al., 2006). During movement, the inertial force equal to mass times the body's
J.M. Smoliga et al. / Journal of Biomechanics 48 (2015) 1950–1956
1951
Fig. 1. Lower body positive pressure treadmill setup.
acceleration (M a) contributes to the vertical GRF
GRF = M·g − A body ΔP + M·a
(1)
Several studies have reported a significantly reduced GRF with greater unloading during walking and running on LBPPT (Cutuk et al., 2006; Grabowski and Kram, 2008; Hoffman and Donaghe, 2011; Raffalt et al., 2013). Gait alteration has also been noted with LBPPT, such as reduced cadence and shortened stride length (Ruckstuhl et al., 2009). However, it remains unknown how running on LBPPT alters more specific components of running mechanics, such as footstrike kinetics. The paucity of scientific information regarding unweighting on LBPPT severely limits the ability of clinicians and coaches to make informed decisions when designing training and rehabilitation protocols using LBPPT. Therefore, the purpose of this study was to quantify the relationship between unweighting magnitude and biomechanical data during running across multiple speeds using LBPPT. Specifically, we examined regional and whole foot plantar forces and impulse calculated from in-shoe pressure data, as well as changes in relative load distribution across different unweighting settings.
2. Methods Subjects: Ten experienced runners (current or former intercollegiate track athletes) participated in this study (age 25.275.6 yr, height 171.378.2 cm, body mass 63.1710.0 kg). Informed written consent was obtained from each subject in accordance with the protocol approved by the University Institutional Review Board. Procedures: Each individual was equipped with 1.9 mm thick, flexible pressure distribution measurement insoles (pedar-x, novel electronics, inc., St. Paul, MN). The in-shoe pressure distribution technology, equipped with 99 sensors in each insole telemetrically transmits data to a computer via a control unit worn at the waist. Calibration was performed per manufacturer recommendation using a customized device which places known pressures upon the insoles (Ford et al., 2006). The minimum threshold defined for each sensor was 2 N. The appropriate size insole was inserted into the left and right running shoe over the normal insole. Loose cables were properly fastened to the subject to assure comfort and prevent cable movement during testing. A brief acclimation period was used to assure proper insole fit. Prior to each running bout, the subject completely unweighted each foot to unload the pressure on the sensors. Subjects did not adjust their footwear throughout the data collection procedure. Subjects ran on a LBPPT (P200, Alter-G, Fremont, CA) at three different speeds and five different body weight percentages (BWSet). Prior to data collection, subjects performed a 5 min self-selected warm-up on the LBPPT. Running speeds were 100%, 120%, and 140% of each individual's self-reported pace during a typical “easy” training run. BWSet were 20%, 40%, 60%, 80%, and 100% of body weight. The order of running speeds was randomized, and the order of BWSet was randomized within each running speed. For each condition, the treadmill settings were adjusted and subjects were allowed a 10–20 s acclimation period before data were collected for 20 s. Subjects were provided a brief rest (approximately two minutes) between each speed. Data analysis: Regional analysis was performed looking at nine individual masks (M1–M9) based on percentage of foot length and width (Ford et al., 2006). Specifically, the lateral (M1) and medial heel (M2), medial (M3) and lateral midfoot (M4), medial (M5), central (M6), and lateral forefoot (M7), hallux (M8), and the lesser toes (M9) were identified (Fig. 2).
The results obtained from individual analysis of the nine masked regions suggested that LBPPT could influence the distribution between forefoot and rear foot relative loading. Therefore, rearfoot was defined as the combination of the medial and lateral rearfoot [M1 þ M2], and forefoot was defined as the combination of forefoot and toes [M5 þM6 þ M7þ M8 þM9]. To determine the distribution between forefoot and rearfoot relative loading, rearfoot relative loading was subtracted from forefoot relative loading, such that a greater value represented a greater forefoot relative load. The first 10 consecutive individual steps from each 20 s trial were identified and exported for further analysis in Matlab (version R2012B, The Mathworks Inc., Natick, MA). Maximum force (FMax) and impulse were computed for each step bilaterally. FMax was normalized to body weight. Aggregate data from 10 strides were computed for each variable bilaterally. Relative load (RL) was determined by dividing each region impulse by the whole foot impulse (Ford et al., 2006). 2.1. Statistical analysis All statistical analyses were performed in SPSS v21.0 and a priori statistical significance was set using α ¼ 0.05. Preliminary analysis using a Kolmogorov–Smirnov test revealed all variables were normally distributed (p40.05 for each), and therefore parametric statistics were used for all analyses. Preliminary analysis using a repeated measures ANOVA comparing side (left versus right) as a repeated factor revealed there were no significant differences between data collected from the left shoe or the right shoe. Therefore, to reduce redundancy, only data from the right foot were used for analysis. Linear mixed effect models using restricted maximum likelihood estimation and an autoregressive repeated moving average measures covariance structure were used to determine how in-shoe loading parameters varied across running speeds and treadmill BWSet. Models were constructed using a variety of repeated measures covariance structures, the model with the lowest Akaike's Information Criteria score was selected as the model of best fit. Subject was included as a random effect for both models. Running speed (100%, 120%, and 140%) and BWSet (20%, 40%, 60%, 80% and 100%) served as fixed effects. For whole foot parameters, fixed effects (Speed and BWSet) were modeled as continuous variables, and the two-way interaction was not included in the model. This approach allowed for the development of prediction equations similar to those produced from a standard linear regression. Percentages were entered as integers (e.g., for 100%, 100 is used rather than 1.0). For regional foot parameters, fixed effects were modeled as categorical variables to allow straightforward interpretation of the results across the multiple masked regions examined. For individual masked regions, preliminary analysis revealed no two-way interactions between Speed and BWSet, and thus these data are only reported for the 100% speed for simplicity’s sake. Of the 150 trials collected (3 intensities 5 treadmill body weight percentage settings 10 subjects), five trials were missing due to technical errors, and thus a total of 145 trials were analyzed. The linear mixed effects model accommodates for missing data without excluding the entire data set obtained from a given subject or condition.
3. Results Mean running speeds across test subjects were 100% selfselected¼3.5 7 0.5 m/s; 120% self-selected ¼4.2 7 0.6 m/s; 140% self-selected ¼ 4.9 70.7 m/s. In-shoe plantar pressure data from one representative subject across all five conditions are presented in Figs. 3 and 4.
1952
J.M. Smoliga et al. / Journal of Biomechanics 48 (2015) 1950–1956
presented below and parameter estimates of the model are presented in Table 1
3.1. Whole foot loading 3.1.1. Maximum in-shoe plantar force The relationship between mean in-shoe maximum plantar force, running speed, and BWSet from all subjects is displayed in Fig. 5. Linear mixed effect modeling revealed significant fixed effects for BWSet [F(126.778) ¼326.107, p o0.001], running speed [F(1130.005)¼ 44.435; p o0.001], and intercept [F(156.192) ¼ 17.635, p o0.001]. The relationship is presented in Eq. (2), with full model parameters presented in Table 1.
FMax = (0.015· BWSet ) + (0.004· Speed) + 0.522
(2)
These data are interpreted as for a linear regression model, such that for every one unit increase in treadmill BWSet, there is a 1.5% increase in maximum force (normalized to body weight) at a given speed. For instance, while using a treadmill BWset of 50% and running at 100% of self-selected training pace would be: 0.015 (50) þ0.004 (100) þ0.522¼ 1.672 times body weight Thus, to prescribe a desired percentage of maximum force (FDesired), treadmill BWSet can be computed, by rearranging the equation to solve for BWSet:
BWSet = {[(0.004· Speed + 2.022)·FDesired ÷ 100] – (0.004· Speed) –0.522} ÷ 0.015
(3)
Given the self-selected running speed parameter contributes little to the equation and is somewhat subjective, a constant based on the mean running speed tested (120% self-selected speed) may be substituted for this parameter to yield a simplified equation:
BWSet = 166.8·FDesired ÷ 100–66.8
(4)
For instance, if one desired to run at 70% FMax, Eq. (3) would yield 51.6% at 100% self-selected speed and 48.4% at 140% selfselected speed, whereas Eq. (4) yields 50% treadmill body weight percentage setting regardless of speed. Given that the lowest possible BWSet on LBPP is 20%, the limits of FDesired are approximately 53.5% to 100% of unsupported running.
Impulse = 0.002· BWSet –0.001· Speed + 0.133
(5)
Following a similar approach to that for FDesired, a desired percentage of impulse (ImpulseDesired) can be prescribed by rearranging the equation to solve for BWSet (Eq. (6)), and this can be further simplified based on 120% of self-selected speed (Eq. (7))
BWSet = {(0.333–0.001· Speed) ·(ImpulseDesired ÷ 100) + 0.001·Speed–0.133} ÷ 0.002
(6)
BWSet = 106.5· ImpulseDesired ÷ 100–6.5
(7)
Given that the lowest possible BWSet on LBPP is 20%, the limits of ImpulseDesired are approximately 35% to 100% of unsupported running.
3.2. Regional loading 3.2.1. Maximum force for individual masks A significant (po0.05) main effect for BWSet was found for Fmax in all nine foot regions (Fig. 7). Significant differences were observed between most BWSet. The only comparisons which were not significantly different for Fmax were for 80% versus 100% in masked regions 1%, 2%, 4%, and 9%, as well as for 20% versus 40% BWSet for region 8. A full matrix of pairwise comparisons between each BWSet for each mask is presented in online supporting Table 1.
3.1.2. In-shoe plantar impulse The relationship between in-shoe impulse, running speed, and BWSet across all subjects is displayed in Fig. 6. Linear mixed effects modeling revealed significant fixed effects for BWSet [F(129.092)¼ 850.136, po0.001], running speed [F(1132.434)¼80.590; po0.001], and intercept [F(133.247)¼93.545, po0.001]. The equation is
3.2.2. Relative load for individual masks Significant main (p o0.05) effects of BWSet were found for relative load in all regions (Fig. 8). Pairwise comparisons revealed differences in relative loading at the lowest BWSet (20%) compared to all other BWSet in M1, M2, M3, M4, M8 and M9. For M3, the only BWSet comparison that was not significantly different was 80% versus 100%. For M1 and M6, 60% and 80% were significantly different, but other BWSet comparisons were not. Additionally, there was a significant difference in relative load between 40% and 100% BWSet. A full matrix of pairwise comparisons between each BWSet for each mask is presented in online supporting Table 2.
Fig. 2. Regional mask configuration.
3.2.3. Forefoot and rearfoot relative load distribution There was a significant (po0.001) effect for BWSet for forefoot to rear foot relative load distribution (Fig. 9). Pairwise comparisons revealed significant (po0.05) differences between all BWSet comparisons, except for 80% versus 100% (p¼ 0.466). Speed (p¼0.060) and the interaction between Speed BWSet (p¼0.999) were not significant.
Fig. 3. Heat map of in-shoe plantar pressure at each body weight for the fast speed condition from one representative subject.
J.M. Smoliga et al. / Journal of Biomechanics 48 (2015) 1950–1956
1953
Fig. 4. In-shoe force–time curves at each body weight at the fast speed condition from one representative subject. Lines represent mean data and color fill represents standard deviations within one subject. Force was greatest at 100% BWSet and reduced with each subsequent decrease in BWSet. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)
Table 1 Linear mixed effect model parameters for maximum force and force–time impulse using body weight percentage setting as a continuous variable.
4. Discussion This study determined the relationship between running speed and programmed body weight percentage on in-shoe whole foot and regional biomechanical parameters during running on LBPPT. The results indicate that maximum in-shoe forces and impulse are linearly related to both speed and body weight percentage, but that relative load on different regions of the foot is not altered
p-Value
0.522
(0.273, 0.771)
o 0.001
Body weight Running speed
0.015 0.004
(0.013, 0.016) (0.003, 0.006)
o 0.001 o 0.001
Intercept Body weight Running speed
0.133 0.002 0.001
(0.105, 0.161) (0.002, 0.003) ( 0.001, 0.000)
o 0.001 o 0.001 o 0.001
Model parameter
Maximum force
Intercept
Impulse
Fig. 5. Relationship between in-shoe maximum plantar force, running speed, and treadmill body weight setting. For each body weight setting, the three different speeds are slightly horizontally offset for clarity. Points represent means and bars represent standard errors.
95% C.I.
Variable
Parameter estimate
uniformly during unweighting. While maximum force and impulse were generally reduced across all regions of the foot, it was disproportionately reduced more in the rearfoot, indicating a transition towards a forefoot/toe strike. Thus, athletes with a forefoot injury (e.g., metatarsal stress fracture) must be conscious of their training volume while running on LBPPT, as the reduced perception of overall impact may provide a false sense of security while the forefoot continues to undergo considerable loading. Further, non-uniform changes in Fmax (Fig. 7) and relative load distribution across the masked regions (Figs. 8–9) indicates that LBPPT alters running mechanics in a complex manner. Nonuniform changes in in-shoe pressure were also reported by Flynn et al. (1997), who found that unweighting using a harness system had the least effect on peak pressure of the medial forefoot during running at 2.7 m/s. While this seems consistent with the present results, direct comparison between studies is not possible, given that we focused on regional force and force–time integral, as these measurements would more likely to be related to musculoskeletal
1954
J.M. Smoliga et al. / Journal of Biomechanics 48 (2015) 1950–1956
loading, rather than localized peak pressure, which may be more relevant to skin ulceration. Maximum in-shoe forces with and without unloading in the current study are generally comparable to previously reported values (Cavanagh et al., 2009; Flynn et al., 1997; Genc et al., 2010; Kluitenberg et al., 2012). Within a given BWSet, faster running speeds were associated with greater Fmax, which is consistent with the literature (Grabowski and Kram, 2008; Hoffman and Donaghe, 2011; Raffalt et al., 2013), and also decreased whole foot impulse. Decreased impulse may result from decreased ground contact time, which has been previously observed during unweighted running (Chang et al., 2000; Millslagle et al., 2006; Raffalt et al., 2013), though this change has not been universal and may be dependent on the protocol used (Sainton et al., in press). Although lower impulse with each individual footstrike may be beneficial in injury prevention, faster running speeds are associated with greater stride rate (Raffalt et al., 2013), and therefore total load may remain unchanged or increase. Collectively, the data indicate that running at 80% BWSet does not substantially alter in-shoe regional plantar loading compared to unsupported running (100% BWSet). This has significant clinical implications, such that Fmax and
impulse can be reduced at this level of unweighting without causing major alterations in running mechanics (i.e., changes in relative load distribution). Thus, at 80% BWSet runners with an injury may be able to continue receiving the cardiometabolic benefits of their normal training volume while experiencing slightly reduced musculoskeletal loading, without major neuromechanical alterations. However, it is possible that the minimal differences between 80% and 100% BWSet may be related to some degree of support being provided even at the 100% BWSet level, likely due to a small amount of positive pressure necessary to keep the chamber inflated (McNeill et al., in press). It should be noted that the values for BWSet are somewhat misleading, which can cause confusion. For instance, running at BWSet of 20% does not produce a maximum force equivalent to 20% of that running without support. Using the self-selected running speed as a reference point, the predicted maximum force at BWSet of 20% is 1.20 times body weight. Comparably, BWSet of 100% yields a predicted maximum 2.32 times body weight. Thus, 20% BWSet results in approximately 52% of maximum force experienced relative to unsupported running (1.20C2.32¼0.517). Therefore, data from this study indicate that the body weight percentage settings on the LBPPT do not correspond to the percentage of maximum force, but rather the percentage of maximum force beyond an individual's body weight. Using the settings from the previous example, at a body weight percentage setting of 100%, a person is landing with 2.32 times body weight, or 1.32 above body weight, while at the 20% BWSet a person is landing at 0.20 above body weight. Indeed, 0.20/1.32 equals 15% (approximately 20%). The 5% discrepancy noted at 20% BWSet during running is consistent with a report that standing weight on LBPPT has about 7% error, which is magnified at the extremes (McNeill et al., in press). While linear statistics yielded significant results in this study, it is likely that the relationship between measured and predicted values becomes nonlinear at BWSet extremes, which explains the presence of an intercept as a “correction factor” in the equations developed. For instance, a theoretical BWSet of 0% should yield no impact forces at any speed, but the prediction equations yield a positive value near body weight. Thus, LBPP prediction equations should only be considered for the BWSet range used to compute them. 4.1. Clinical applications
Fig. 6. Relationship between in-shoe impulse, running speed, and treadmill body weight setting. For each body weight setting, the three different speeds are slightly horizontally offset for esthetic purposes. Points represent means and bars represent standard errors.
The results of the present study have widespread generalizable clinical application, given the commercial availability of LBPPT. The clinical utility of LBPPT in reducing musculoskeletal load becomes especially apparent when one compares the magnitude of changes observed at different BWSet to those experienced at different
Fig. 7. Maximum in-shoe plantar force for each foot region at the slow speed. Only the 100% self-selected speed is presented. Bars represent standard errors.
J.M. Smoliga et al. / Journal of Biomechanics 48 (2015) 1950–1956
1955
Fig. 8. Percent relative load for each foot region. Bars represent standard errors.
Fig. 9. Change in relative load distribution with unweighing across three speeds. Points represent means and bars represent standard errors.
running speeds (Warden et al., 2014). The wide range of speeds used in the study were intended to be representative of the range of speeds a runner would typically use in a structured training program (i.e., easy running through high-intensity interval training). Both speed and BWSet were demonstrated to independently have significant effects on Fmax and impulse. While rehabilitation from running injuries often focuses on lower-intensity running to reduce musculoskeletal loading (Warden et al., 2014) it is clear that athletes can also reduce musculoskeletal loading while performing relatively high-speed running on LBPPT. For instance, impulse imposed upon the foot at 100% self-selected speed at 100% BWSet is approximately equivalent to that at 140% selfselected speed at 80% BWSet (Fig. 6). Though LBPPT unweighting does reduce metabolic stimulus (Kline et al., 2015; Raffalt et al., 2013), higher speed running may be more neuromechanically equivalent to that experienced during training, and thus, LBPPT has potential to allow injured athletes to maintain their typical training structure through rehabilitation. Eqs. (4) and (6) will allow clinicians to develop patient-specific protocols which achieve the desired maximal forces and impulses. To facilitate the use of these equations, a Microsoft Excel worksheet which automatically computes these values and a table of FMax and impulse at various BWSet are available as online supporting material.
Although such prescriptions are desirable from a clinical standpoint, there is a lack of evidence-based recommendations for specific unweighting settings, and therefore the desired magnitude of unweighting is based on clinical judgment. Likewise, researchers can use these equations to study the effects of different types of impact forces. While maximum force and impulse are closely related, one may use different combinations of running speed and body weight percentage to achieve different loading effects. Both peak force and impulse have been identified as potential risk factors for common overuse running related injuries and both measures were reduced in patients following gait retraining (Samaan et al., 2014). Gait retraining with the aid of LBPPT in certain patients may also be beneficial and should be further investigated. The equation chosen for exercise prescription for an individual will ultimately depend on specific clinical needs. While the unweighting experienced in LBPPT may be advantageous for rehabilitation of various injuries, changes in in-shoe regional plantar loading suggest that clinicians use some caution for exercise prescription. It is not yet known if regular training with these altered mechanics ultimately influences neuromechanics during overground running (i.e., unintentional gait retraining). While alterations in relative load distribution occur through the continuum of BWSet examined, it is most extreme at 20%. Thus, training using the 20% setting may not provide an appropriate neuromechanical stimulus that translates into overground running. 4.2. Limitations There are some limitations to this study which the authors have attempted to address. It is possible that familiarization with LBPPT may influence the results. Most, but not all, of the subjects had previous LBPPT experience, and inexperienced subjects did not anecdotally report any familiarization issues. While fatigue could theoretically affect the results, these trained subjects were unlikely to fatigue during short duration of this protocol, and randomization likely eliminated order effects. As demonstrated in Fig. 4, the time of the maximum impact also may shift with different BWSet, occurring during propulsion at higher BWSet, and during breaking at lower BWSet, which can make interpretation of this data difficult. However, comparison at similar stride-duration normalized time points also becomes difficult, as the absolute stride duration changes at different BWSet (Raffalt et al., 2013). This does not invalidate the data, but rather emphasizes that LBPPT changes biomechanics sufficiently to complicate comparisons between BWSet, which emphasizes the
1956
J.M. Smoliga et al. / Journal of Biomechanics 48 (2015) 1950–1956
need for appropriate clinical translation. Although the sample size was relatively small, visual inspection of individual plots revealed results were generally consistent between individuals.
5. Conclusions In conclusion, the present data indicate that LBPPT significantly reduces in-shoe maximum forces and impulse during running across multiple speeds, though there is a disconnect between treadmill settings and actual unloading. Speed does independently influence whole foot Fmax and impulse; however a relatively small magnitude of unweighting while running on LBPPT reduces these in-shoe loading parameters to a similar magnitude attained by running slower. This may be especially relevant for injury-prone or already injured athletes, as LPPPT may allow well-trained individuals to experience the neuromechanical and metabolic advantages of high speed running while reducing musculoskeletal loading. Differences in the magnitude of changes in Fmax and impulse exist between different regions of the foot, such that the relative load distribution is shifted towards the forefoot with more unweighting. It appears that 80% BWSet may provide a balance between reducing Fmax and impulse, while minimally disrupting in-shoe regional loading. Conversely, 20% BWSet appears to cause major changes in multiple loading parameters and may be of limited clinical utility. These findings can be used for exercise prescription in rehabilitation and sports performance settings, and may also be used by researchers to study the effects of musculoskeletal unloading during exercise.
Conflict of interest statement The authors have no conflicts of interest, financial or otherwise, to declare.
Acknowledgments This was an investigator-initiated study and no funding or equipment was provided for this study.
Appendix A. Supplementary material Supplementary material associated with this article can be found in the online version at http://dx.doi.org/10.1016/j.jbiomech. 2015.04.009.
References Cavanagh, P.R., Gopalakrishnan, R., Rice, A.J., Genc, K.O., Maender, C.C., Nystrom, P.G., Johnson, M.J., Kuklis, M.M., Humphreys, B.T., 2009. An ambulatory biomechanical data collection system for use in space: design and validation. Aviat. Space Environ. Med. 80, 870–881. Chang, Y.H., Huang, H.W., Hamerski, C.M., Kram, R., 2000. The independent effects of gravity and inertia on running mechanics. J. Exp. Biol. 203, 229–238. Cutuk, A., Groppo, E.R., Quigley, E.J., White, K.W., Pedowitz, R.A., Hargens, A.R., 2006. Ambulation in simulated fractional gravity using lower body positive pressure: cardiovascular safety and gait analyses. J. Appl. Physiol. 101, 771–777 (1985). Flynn, T.W., Canavan, P.K., Cavanagh, P.R., Chiang, J.H., 1997. Plantar pressure reduction in an incremental weight-bearing system. Phys. Ther. 77, 410–416. Ford, K.R., Manson, N.A., Evans, B.J., Myer, G.D., Gwin, R.C., Heidt Jr., R.S., Hewett, T.E., 2006. Comparison of in-shoe foot loading patterns on natural grass and synthetic turf. J. Sci. Med. Sport 9, 433–440. Genc, K.O., Gopalakrishnan, R., Kuklis, M.M., Maender, C.C., Rice, A.J., Bowersox, K.D., Cavanagh, P.R., 2010. Foot forces during exercise on the international space station. J. Biomech. 43, 3020–3027. Grabowski, A.M., Kram, R., 2008. Effects of velocity and weight support on ground reaction forces and metabolic power during running. J. Appl. Biomech. 24, 288–297. Hoffman, M.D., Donaghe, H.E., 2011. Physiological responses to body weight–supported treadmill exercise in healthy adults. Arch. Phys. Med. Rehabil. 92, 960–966. Kline, J.R., Raab, S., Coast, J.R., Bounds, R., McNeill, D., de Heer, H.D., 2015. A conversion table for running on Lower Body Positive Pressure (LBPP) treadmills. J. Strength Cond. Res. 29, 863–868. Kluitenberg, B., Bredeweg, S.W., Zijlstra, S., Zijlstra, W., Buist, I., 2012. Comparison of vertical ground reaction forces during overground and treadmill running. A validation study. BMC Musculoskelet. Disord. 13, 235. McNeill, D.K., de Heer, H.D., Bounds, R.G., Coast, J.R., 2015. Accuracy of unloading with the Anti-Gravitys Treadmill. J. Strength Cond. Res. (in press) Millslagle, D., Levy, M., Matack, N., 2006. Kinematic assessment of treadmill running using different body-weight support harnesses. Percept. Mot. Skills 103, 607–618. Newman, D.J., Alexander, H.L., Webbon, B.W., 1994. Energetics and mechanics for partial gravity locomotion. Aviat. Space Environ. Med. 65, 815–823. Raffalt, P.C., Hovgaard-Hansen, L., Jensen, B.R., 2013. Running on a lower-body positive pressure treadmill: VO2max, respiratory response, and vertical ground reaction force. Res. Q. Exerc. Sport 84, 213–222. Ruckstuhl, H., Kho, J., Weed, M., Wilkinson, M.W., Hargens, A.R., 2009. Comparing two devices of suspended treadmill walking by varying body unloading and froude number. Gait Posture 30, 446–451. Sainton, P., Nicol, C., Cabri, J., Barthelemy-Montfort, J., Berton, E., Chavet, P., 2015. Influence of short-term unweighing and reloading on running kinetics and muscle activity. Eur. J. Appl. Physiol. (in press) Samaan, C.D., Rainbow, M.J., Davis, I.S., 2014. Reduction in ground reaction force variables with instructed barefoot running. J. Sport Health Sci. 3, 143–151. Warden, S.J., Davis, I.S., Fredericson, M., 2014. Management and prevention of bone stress injuries in long-distance runners. J. Orthop. Sports Phys. Ther. 44, 749–765.
Contents lists available at ScienceDirect
Journal of Biomechanics journal homepage: www.elsevier.com/locate/jbiomech www.JBiomech.com
Effects of unweighting and speed on in-shoe regional loading during running on a lower body positive pressure treadmill James M. Smoliga a,n, Leah Anne Wirfel b, Danielle Paul b, Mary Doarnberger b, Kevin R. Ford a a b
Human Biomechanics and Physiology Laboratory, Department of Physical Therapy, High Point University, High Point, NC 27268, USA Department of Exercise Science, High Point University, High Point, NC 27409, USA
art ic l e i nf o
a b s t r a c t
Article history: Accepted 3 April 2015
The purpose of this study was to determine how unweighted running on a lower body positive pressure treadmill (LBPPT) modifies in-shoe regional loading. Ten experienced runners were fit with pressure distribution measurement insoles and ran at 100%, 120%, and 140% of self-reported easy training pace on a LBPPT at 20%, 40%, 60%, 80%, and 100% body weight percentage settings (BWSet). Speeds and BWSet were in random order. A linear mixed effect model (po0.05 significance level) was used to compare differences in whole foot and regional maximum in-shoe plantar force (FMAX), impulse, and relative load distribution across speeds and BWSet. There were significant main effects (po 0.001) for running speed and BWSet for whole foot Fmax and impulse. The model revealed 1.4% and 0.24% increases in whole foot FMAX (times body weight) and impulse, respectively, for every unit increase in body weight percentage. There was a significant main effect for BWSet on Fmax and relative load (p o0.05) for each of the nine foot regions examined, though four regions were not different between 80% and 100% BWSet. There was a significant (po 0.001) main effect for BWSet on forefoot to rear foot relative load. Linear relationships were found between increases in BWSet and increases in-shoe Fmax and impulse, resulting from regional changes in foot pressure which represent a shift towards forefoot loading, most evident o 80% BWSet. Estimating in-shoe regional loading parameters may be useful during rehabilitation and training to appropriately prescribe specific speed and body weight levels, without exceeding certain critical peak force levels while running. & 2015 Elsevier Ltd. All rights reserved.
Keywords: Running Unweighting Unloading Treadmill Overuse injury rehabilitation Stress fracture
1. Introduction Applied biomechanics experts regularly adopt new technology to optimize musculoskeletal rehabilitation and maximize athletic performance. While various methods to reduce musculoskeletal loading during ambulation are available, such as underwater treadmills and harness systems (Newman et al., 1994), lower body positive pressure treadmills (LBPPT) provide a novel method of unweighted exercise and are becoming more readily available in rehabilitation and sports performance settings. Reduced foot and overall loading on the lower extremity may be beneficial to those rehabilitating from certain overuse injuries (Samaan et al., 2014; Warden et al., 2014). For instance, a graded return to running program would be designed to not only reduce the maximum ground reaction force (GRF) but also the total force time-integral during each stance phase. However, n Correspondence to: Associate Professor of Physiology, Department of Physical Therapy 833 Montlieu Avenue, High Point University, High Point, NC 27268. Tel.: þ 1 336 841 9480; fax: þ1 336 888 6394. E-mail address: [email protected] (J.M. Smoliga).
http://dx.doi.org/10.1016/j.jbiomech.2015.04.009 0021-9290/& 2015 Elsevier Ltd. All rights reserved.
certain individuals that may benefit from reduced plantar loading during rehabilitation, such as those returning from a metatarsal stress fracture, may alter running mechanics and potentially not reduce explicit plantar regions. Flynn et al. (1997) identified a nonconsistent reduction of regional peak pressure during running using an unweighted harness system. However, the force and time integral were not analyzed regionally, nor were measurements across multiple running speeds or unweighting magnitudes reported. Therefore, to ensure appropriate use during rehabilitation and training, it is important to understand specific alterations in regional plantar loading under various conditions. When exercising in LBPPT, airflow is directed into a capsule surrounding a treadmill and lower extremities of the athlete or patient (Fig. 1). Air pressure is increased within the capsule which applies an upward directed vertical force on the user. Therefore, the net GRF (Eq. (1)) is reduced as the downward gravitational force (M g, mass gravity) by the upward directed force from the air pressure differential (ΔP) multiplied by the cross-sectional area of the lower body and seal (Abody) (Cutuk et al., 2006). During movement, the inertial force equal to mass times the body's
J.M. Smoliga et al. / Journal of Biomechanics 48 (2015) 1950–1956
1951
Fig. 1. Lower body positive pressure treadmill setup.
acceleration (M a) contributes to the vertical GRF
GRF = M·g − A body ΔP + M·a
(1)
Several studies have reported a significantly reduced GRF with greater unloading during walking and running on LBPPT (Cutuk et al., 2006; Grabowski and Kram, 2008; Hoffman and Donaghe, 2011; Raffalt et al., 2013). Gait alteration has also been noted with LBPPT, such as reduced cadence and shortened stride length (Ruckstuhl et al., 2009). However, it remains unknown how running on LBPPT alters more specific components of running mechanics, such as footstrike kinetics. The paucity of scientific information regarding unweighting on LBPPT severely limits the ability of clinicians and coaches to make informed decisions when designing training and rehabilitation protocols using LBPPT. Therefore, the purpose of this study was to quantify the relationship between unweighting magnitude and biomechanical data during running across multiple speeds using LBPPT. Specifically, we examined regional and whole foot plantar forces and impulse calculated from in-shoe pressure data, as well as changes in relative load distribution across different unweighting settings.
2. Methods Subjects: Ten experienced runners (current or former intercollegiate track athletes) participated in this study (age 25.275.6 yr, height 171.378.2 cm, body mass 63.1710.0 kg). Informed written consent was obtained from each subject in accordance with the protocol approved by the University Institutional Review Board. Procedures: Each individual was equipped with 1.9 mm thick, flexible pressure distribution measurement insoles (pedar-x, novel electronics, inc., St. Paul, MN). The in-shoe pressure distribution technology, equipped with 99 sensors in each insole telemetrically transmits data to a computer via a control unit worn at the waist. Calibration was performed per manufacturer recommendation using a customized device which places known pressures upon the insoles (Ford et al., 2006). The minimum threshold defined for each sensor was 2 N. The appropriate size insole was inserted into the left and right running shoe over the normal insole. Loose cables were properly fastened to the subject to assure comfort and prevent cable movement during testing. A brief acclimation period was used to assure proper insole fit. Prior to each running bout, the subject completely unweighted each foot to unload the pressure on the sensors. Subjects did not adjust their footwear throughout the data collection procedure. Subjects ran on a LBPPT (P200, Alter-G, Fremont, CA) at three different speeds and five different body weight percentages (BWSet). Prior to data collection, subjects performed a 5 min self-selected warm-up on the LBPPT. Running speeds were 100%, 120%, and 140% of each individual's self-reported pace during a typical “easy” training run. BWSet were 20%, 40%, 60%, 80%, and 100% of body weight. The order of running speeds was randomized, and the order of BWSet was randomized within each running speed. For each condition, the treadmill settings were adjusted and subjects were allowed a 10–20 s acclimation period before data were collected for 20 s. Subjects were provided a brief rest (approximately two minutes) between each speed. Data analysis: Regional analysis was performed looking at nine individual masks (M1–M9) based on percentage of foot length and width (Ford et al., 2006). Specifically, the lateral (M1) and medial heel (M2), medial (M3) and lateral midfoot (M4), medial (M5), central (M6), and lateral forefoot (M7), hallux (M8), and the lesser toes (M9) were identified (Fig. 2).
The results obtained from individual analysis of the nine masked regions suggested that LBPPT could influence the distribution between forefoot and rear foot relative loading. Therefore, rearfoot was defined as the combination of the medial and lateral rearfoot [M1 þ M2], and forefoot was defined as the combination of forefoot and toes [M5 þM6 þ M7þ M8 þM9]. To determine the distribution between forefoot and rearfoot relative loading, rearfoot relative loading was subtracted from forefoot relative loading, such that a greater value represented a greater forefoot relative load. The first 10 consecutive individual steps from each 20 s trial were identified and exported for further analysis in Matlab (version R2012B, The Mathworks Inc., Natick, MA). Maximum force (FMax) and impulse were computed for each step bilaterally. FMax was normalized to body weight. Aggregate data from 10 strides were computed for each variable bilaterally. Relative load (RL) was determined by dividing each region impulse by the whole foot impulse (Ford et al., 2006). 2.1. Statistical analysis All statistical analyses were performed in SPSS v21.0 and a priori statistical significance was set using α ¼ 0.05. Preliminary analysis using a Kolmogorov–Smirnov test revealed all variables were normally distributed (p40.05 for each), and therefore parametric statistics were used for all analyses. Preliminary analysis using a repeated measures ANOVA comparing side (left versus right) as a repeated factor revealed there were no significant differences between data collected from the left shoe or the right shoe. Therefore, to reduce redundancy, only data from the right foot were used for analysis. Linear mixed effect models using restricted maximum likelihood estimation and an autoregressive repeated moving average measures covariance structure were used to determine how in-shoe loading parameters varied across running speeds and treadmill BWSet. Models were constructed using a variety of repeated measures covariance structures, the model with the lowest Akaike's Information Criteria score was selected as the model of best fit. Subject was included as a random effect for both models. Running speed (100%, 120%, and 140%) and BWSet (20%, 40%, 60%, 80% and 100%) served as fixed effects. For whole foot parameters, fixed effects (Speed and BWSet) were modeled as continuous variables, and the two-way interaction was not included in the model. This approach allowed for the development of prediction equations similar to those produced from a standard linear regression. Percentages were entered as integers (e.g., for 100%, 100 is used rather than 1.0). For regional foot parameters, fixed effects were modeled as categorical variables to allow straightforward interpretation of the results across the multiple masked regions examined. For individual masked regions, preliminary analysis revealed no two-way interactions between Speed and BWSet, and thus these data are only reported for the 100% speed for simplicity’s sake. Of the 150 trials collected (3 intensities 5 treadmill body weight percentage settings 10 subjects), five trials were missing due to technical errors, and thus a total of 145 trials were analyzed. The linear mixed effects model accommodates for missing data without excluding the entire data set obtained from a given subject or condition.
3. Results Mean running speeds across test subjects were 100% selfselected¼3.5 7 0.5 m/s; 120% self-selected ¼4.2 7 0.6 m/s; 140% self-selected ¼ 4.9 70.7 m/s. In-shoe plantar pressure data from one representative subject across all five conditions are presented in Figs. 3 and 4.
1952
J.M. Smoliga et al. / Journal of Biomechanics 48 (2015) 1950–1956
presented below and parameter estimates of the model are presented in Table 1
3.1. Whole foot loading 3.1.1. Maximum in-shoe plantar force The relationship between mean in-shoe maximum plantar force, running speed, and BWSet from all subjects is displayed in Fig. 5. Linear mixed effect modeling revealed significant fixed effects for BWSet [F(126.778) ¼326.107, p o0.001], running speed [F(1130.005)¼ 44.435; p o0.001], and intercept [F(156.192) ¼ 17.635, p o0.001]. The relationship is presented in Eq. (2), with full model parameters presented in Table 1.
FMax = (0.015· BWSet ) + (0.004· Speed) + 0.522
(2)
These data are interpreted as for a linear regression model, such that for every one unit increase in treadmill BWSet, there is a 1.5% increase in maximum force (normalized to body weight) at a given speed. For instance, while using a treadmill BWset of 50% and running at 100% of self-selected training pace would be: 0.015 (50) þ0.004 (100) þ0.522¼ 1.672 times body weight Thus, to prescribe a desired percentage of maximum force (FDesired), treadmill BWSet can be computed, by rearranging the equation to solve for BWSet:
BWSet = {[(0.004· Speed + 2.022)·FDesired ÷ 100] – (0.004· Speed) –0.522} ÷ 0.015
(3)
Given the self-selected running speed parameter contributes little to the equation and is somewhat subjective, a constant based on the mean running speed tested (120% self-selected speed) may be substituted for this parameter to yield a simplified equation:
BWSet = 166.8·FDesired ÷ 100–66.8
(4)
For instance, if one desired to run at 70% FMax, Eq. (3) would yield 51.6% at 100% self-selected speed and 48.4% at 140% selfselected speed, whereas Eq. (4) yields 50% treadmill body weight percentage setting regardless of speed. Given that the lowest possible BWSet on LBPP is 20%, the limits of FDesired are approximately 53.5% to 100% of unsupported running.
Impulse = 0.002· BWSet –0.001· Speed + 0.133
(5)
Following a similar approach to that for FDesired, a desired percentage of impulse (ImpulseDesired) can be prescribed by rearranging the equation to solve for BWSet (Eq. (6)), and this can be further simplified based on 120% of self-selected speed (Eq. (7))
BWSet = {(0.333–0.001· Speed) ·(ImpulseDesired ÷ 100) + 0.001·Speed–0.133} ÷ 0.002
(6)
BWSet = 106.5· ImpulseDesired ÷ 100–6.5
(7)
Given that the lowest possible BWSet on LBPP is 20%, the limits of ImpulseDesired are approximately 35% to 100% of unsupported running.
3.2. Regional loading 3.2.1. Maximum force for individual masks A significant (po0.05) main effect for BWSet was found for Fmax in all nine foot regions (Fig. 7). Significant differences were observed between most BWSet. The only comparisons which were not significantly different for Fmax were for 80% versus 100% in masked regions 1%, 2%, 4%, and 9%, as well as for 20% versus 40% BWSet for region 8. A full matrix of pairwise comparisons between each BWSet for each mask is presented in online supporting Table 1.
3.1.2. In-shoe plantar impulse The relationship between in-shoe impulse, running speed, and BWSet across all subjects is displayed in Fig. 6. Linear mixed effects modeling revealed significant fixed effects for BWSet [F(129.092)¼ 850.136, po0.001], running speed [F(1132.434)¼80.590; po0.001], and intercept [F(133.247)¼93.545, po0.001]. The equation is
3.2.2. Relative load for individual masks Significant main (p o0.05) effects of BWSet were found for relative load in all regions (Fig. 8). Pairwise comparisons revealed differences in relative loading at the lowest BWSet (20%) compared to all other BWSet in M1, M2, M3, M4, M8 and M9. For M3, the only BWSet comparison that was not significantly different was 80% versus 100%. For M1 and M6, 60% and 80% were significantly different, but other BWSet comparisons were not. Additionally, there was a significant difference in relative load between 40% and 100% BWSet. A full matrix of pairwise comparisons between each BWSet for each mask is presented in online supporting Table 2.
Fig. 2. Regional mask configuration.
3.2.3. Forefoot and rearfoot relative load distribution There was a significant (po0.001) effect for BWSet for forefoot to rear foot relative load distribution (Fig. 9). Pairwise comparisons revealed significant (po0.05) differences between all BWSet comparisons, except for 80% versus 100% (p¼ 0.466). Speed (p¼0.060) and the interaction between Speed BWSet (p¼0.999) were not significant.
Fig. 3. Heat map of in-shoe plantar pressure at each body weight for the fast speed condition from one representative subject.
J.M. Smoliga et al. / Journal of Biomechanics 48 (2015) 1950–1956
1953
Fig. 4. In-shoe force–time curves at each body weight at the fast speed condition from one representative subject. Lines represent mean data and color fill represents standard deviations within one subject. Force was greatest at 100% BWSet and reduced with each subsequent decrease in BWSet. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)
Table 1 Linear mixed effect model parameters for maximum force and force–time impulse using body weight percentage setting as a continuous variable.
4. Discussion This study determined the relationship between running speed and programmed body weight percentage on in-shoe whole foot and regional biomechanical parameters during running on LBPPT. The results indicate that maximum in-shoe forces and impulse are linearly related to both speed and body weight percentage, but that relative load on different regions of the foot is not altered
p-Value
0.522
(0.273, 0.771)
o 0.001
Body weight Running speed
0.015 0.004
(0.013, 0.016) (0.003, 0.006)
o 0.001 o 0.001
Intercept Body weight Running speed
0.133 0.002 0.001
(0.105, 0.161) (0.002, 0.003) ( 0.001, 0.000)
o 0.001 o 0.001 o 0.001
Model parameter
Maximum force
Intercept
Impulse
Fig. 5. Relationship between in-shoe maximum plantar force, running speed, and treadmill body weight setting. For each body weight setting, the three different speeds are slightly horizontally offset for clarity. Points represent means and bars represent standard errors.
95% C.I.
Variable
Parameter estimate
uniformly during unweighting. While maximum force and impulse were generally reduced across all regions of the foot, it was disproportionately reduced more in the rearfoot, indicating a transition towards a forefoot/toe strike. Thus, athletes with a forefoot injury (e.g., metatarsal stress fracture) must be conscious of their training volume while running on LBPPT, as the reduced perception of overall impact may provide a false sense of security while the forefoot continues to undergo considerable loading. Further, non-uniform changes in Fmax (Fig. 7) and relative load distribution across the masked regions (Figs. 8–9) indicates that LBPPT alters running mechanics in a complex manner. Nonuniform changes in in-shoe pressure were also reported by Flynn et al. (1997), who found that unweighting using a harness system had the least effect on peak pressure of the medial forefoot during running at 2.7 m/s. While this seems consistent with the present results, direct comparison between studies is not possible, given that we focused on regional force and force–time integral, as these measurements would more likely to be related to musculoskeletal
1954
J.M. Smoliga et al. / Journal of Biomechanics 48 (2015) 1950–1956
loading, rather than localized peak pressure, which may be more relevant to skin ulceration. Maximum in-shoe forces with and without unloading in the current study are generally comparable to previously reported values (Cavanagh et al., 2009; Flynn et al., 1997; Genc et al., 2010; Kluitenberg et al., 2012). Within a given BWSet, faster running speeds were associated with greater Fmax, which is consistent with the literature (Grabowski and Kram, 2008; Hoffman and Donaghe, 2011; Raffalt et al., 2013), and also decreased whole foot impulse. Decreased impulse may result from decreased ground contact time, which has been previously observed during unweighted running (Chang et al., 2000; Millslagle et al., 2006; Raffalt et al., 2013), though this change has not been universal and may be dependent on the protocol used (Sainton et al., in press). Although lower impulse with each individual footstrike may be beneficial in injury prevention, faster running speeds are associated with greater stride rate (Raffalt et al., 2013), and therefore total load may remain unchanged or increase. Collectively, the data indicate that running at 80% BWSet does not substantially alter in-shoe regional plantar loading compared to unsupported running (100% BWSet). This has significant clinical implications, such that Fmax and
impulse can be reduced at this level of unweighting without causing major alterations in running mechanics (i.e., changes in relative load distribution). Thus, at 80% BWSet runners with an injury may be able to continue receiving the cardiometabolic benefits of their normal training volume while experiencing slightly reduced musculoskeletal loading, without major neuromechanical alterations. However, it is possible that the minimal differences between 80% and 100% BWSet may be related to some degree of support being provided even at the 100% BWSet level, likely due to a small amount of positive pressure necessary to keep the chamber inflated (McNeill et al., in press). It should be noted that the values for BWSet are somewhat misleading, which can cause confusion. For instance, running at BWSet of 20% does not produce a maximum force equivalent to 20% of that running without support. Using the self-selected running speed as a reference point, the predicted maximum force at BWSet of 20% is 1.20 times body weight. Comparably, BWSet of 100% yields a predicted maximum 2.32 times body weight. Thus, 20% BWSet results in approximately 52% of maximum force experienced relative to unsupported running (1.20C2.32¼0.517). Therefore, data from this study indicate that the body weight percentage settings on the LBPPT do not correspond to the percentage of maximum force, but rather the percentage of maximum force beyond an individual's body weight. Using the settings from the previous example, at a body weight percentage setting of 100%, a person is landing with 2.32 times body weight, or 1.32 above body weight, while at the 20% BWSet a person is landing at 0.20 above body weight. Indeed, 0.20/1.32 equals 15% (approximately 20%). The 5% discrepancy noted at 20% BWSet during running is consistent with a report that standing weight on LBPPT has about 7% error, which is magnified at the extremes (McNeill et al., in press). While linear statistics yielded significant results in this study, it is likely that the relationship between measured and predicted values becomes nonlinear at BWSet extremes, which explains the presence of an intercept as a “correction factor” in the equations developed. For instance, a theoretical BWSet of 0% should yield no impact forces at any speed, but the prediction equations yield a positive value near body weight. Thus, LBPP prediction equations should only be considered for the BWSet range used to compute them. 4.1. Clinical applications
Fig. 6. Relationship between in-shoe impulse, running speed, and treadmill body weight setting. For each body weight setting, the three different speeds are slightly horizontally offset for esthetic purposes. Points represent means and bars represent standard errors.
The results of the present study have widespread generalizable clinical application, given the commercial availability of LBPPT. The clinical utility of LBPPT in reducing musculoskeletal load becomes especially apparent when one compares the magnitude of changes observed at different BWSet to those experienced at different
Fig. 7. Maximum in-shoe plantar force for each foot region at the slow speed. Only the 100% self-selected speed is presented. Bars represent standard errors.
J.M. Smoliga et al. / Journal of Biomechanics 48 (2015) 1950–1956
1955
Fig. 8. Percent relative load for each foot region. Bars represent standard errors.
Fig. 9. Change in relative load distribution with unweighing across three speeds. Points represent means and bars represent standard errors.
running speeds (Warden et al., 2014). The wide range of speeds used in the study were intended to be representative of the range of speeds a runner would typically use in a structured training program (i.e., easy running through high-intensity interval training). Both speed and BWSet were demonstrated to independently have significant effects on Fmax and impulse. While rehabilitation from running injuries often focuses on lower-intensity running to reduce musculoskeletal loading (Warden et al., 2014) it is clear that athletes can also reduce musculoskeletal loading while performing relatively high-speed running on LBPPT. For instance, impulse imposed upon the foot at 100% self-selected speed at 100% BWSet is approximately equivalent to that at 140% selfselected speed at 80% BWSet (Fig. 6). Though LBPPT unweighting does reduce metabolic stimulus (Kline et al., 2015; Raffalt et al., 2013), higher speed running may be more neuromechanically equivalent to that experienced during training, and thus, LBPPT has potential to allow injured athletes to maintain their typical training structure through rehabilitation. Eqs. (4) and (6) will allow clinicians to develop patient-specific protocols which achieve the desired maximal forces and impulses. To facilitate the use of these equations, a Microsoft Excel worksheet which automatically computes these values and a table of FMax and impulse at various BWSet are available as online supporting material.
Although such prescriptions are desirable from a clinical standpoint, there is a lack of evidence-based recommendations for specific unweighting settings, and therefore the desired magnitude of unweighting is based on clinical judgment. Likewise, researchers can use these equations to study the effects of different types of impact forces. While maximum force and impulse are closely related, one may use different combinations of running speed and body weight percentage to achieve different loading effects. Both peak force and impulse have been identified as potential risk factors for common overuse running related injuries and both measures were reduced in patients following gait retraining (Samaan et al., 2014). Gait retraining with the aid of LBPPT in certain patients may also be beneficial and should be further investigated. The equation chosen for exercise prescription for an individual will ultimately depend on specific clinical needs. While the unweighting experienced in LBPPT may be advantageous for rehabilitation of various injuries, changes in in-shoe regional plantar loading suggest that clinicians use some caution for exercise prescription. It is not yet known if regular training with these altered mechanics ultimately influences neuromechanics during overground running (i.e., unintentional gait retraining). While alterations in relative load distribution occur through the continuum of BWSet examined, it is most extreme at 20%. Thus, training using the 20% setting may not provide an appropriate neuromechanical stimulus that translates into overground running. 4.2. Limitations There are some limitations to this study which the authors have attempted to address. It is possible that familiarization with LBPPT may influence the results. Most, but not all, of the subjects had previous LBPPT experience, and inexperienced subjects did not anecdotally report any familiarization issues. While fatigue could theoretically affect the results, these trained subjects were unlikely to fatigue during short duration of this protocol, and randomization likely eliminated order effects. As demonstrated in Fig. 4, the time of the maximum impact also may shift with different BWSet, occurring during propulsion at higher BWSet, and during breaking at lower BWSet, which can make interpretation of this data difficult. However, comparison at similar stride-duration normalized time points also becomes difficult, as the absolute stride duration changes at different BWSet (Raffalt et al., 2013). This does not invalidate the data, but rather emphasizes that LBPPT changes biomechanics sufficiently to complicate comparisons between BWSet, which emphasizes the
1956
J.M. Smoliga et al. / Journal of Biomechanics 48 (2015) 1950–1956
need for appropriate clinical translation. Although the sample size was relatively small, visual inspection of individual plots revealed results were generally consistent between individuals.
5. Conclusions In conclusion, the present data indicate that LBPPT significantly reduces in-shoe maximum forces and impulse during running across multiple speeds, though there is a disconnect between treadmill settings and actual unloading. Speed does independently influence whole foot Fmax and impulse; however a relatively small magnitude of unweighting while running on LBPPT reduces these in-shoe loading parameters to a similar magnitude attained by running slower. This may be especially relevant for injury-prone or already injured athletes, as LPPPT may allow well-trained individuals to experience the neuromechanical and metabolic advantages of high speed running while reducing musculoskeletal loading. Differences in the magnitude of changes in Fmax and impulse exist between different regions of the foot, such that the relative load distribution is shifted towards the forefoot with more unweighting. It appears that 80% BWSet may provide a balance between reducing Fmax and impulse, while minimally disrupting in-shoe regional loading. Conversely, 20% BWSet appears to cause major changes in multiple loading parameters and may be of limited clinical utility. These findings can be used for exercise prescription in rehabilitation and sports performance settings, and may also be used by researchers to study the effects of musculoskeletal unloading during exercise.
Conflict of interest statement The authors have no conflicts of interest, financial or otherwise, to declare.
Acknowledgments This was an investigator-initiated study and no funding or equipment was provided for this study.
Appendix A. Supplementary material Supplementary material associated with this article can be found in the online version at http://dx.doi.org/10.1016/j.jbiomech. 2015.04.009.
References Cavanagh, P.R., Gopalakrishnan, R., Rice, A.J., Genc, K.O., Maender, C.C., Nystrom, P.G., Johnson, M.J., Kuklis, M.M., Humphreys, B.T., 2009. An ambulatory biomechanical data collection system for use in space: design and validation. Aviat. Space Environ. Med. 80, 870–881. Chang, Y.H., Huang, H.W., Hamerski, C.M., Kram, R., 2000. The independent effects of gravity and inertia on running mechanics. J. Exp. Biol. 203, 229–238. Cutuk, A., Groppo, E.R., Quigley, E.J., White, K.W., Pedowitz, R.A., Hargens, A.R., 2006. Ambulation in simulated fractional gravity using lower body positive pressure: cardiovascular safety and gait analyses. J. Appl. Physiol. 101, 771–777 (1985). Flynn, T.W., Canavan, P.K., Cavanagh, P.R., Chiang, J.H., 1997. Plantar pressure reduction in an incremental weight-bearing system. Phys. Ther. 77, 410–416. Ford, K.R., Manson, N.A., Evans, B.J., Myer, G.D., Gwin, R.C., Heidt Jr., R.S., Hewett, T.E., 2006. Comparison of in-shoe foot loading patterns on natural grass and synthetic turf. J. Sci. Med. Sport 9, 433–440. Genc, K.O., Gopalakrishnan, R., Kuklis, M.M., Maender, C.C., Rice, A.J., Bowersox, K.D., Cavanagh, P.R., 2010. Foot forces during exercise on the international space station. J. Biomech. 43, 3020–3027. Grabowski, A.M., Kram, R., 2008. Effects of velocity and weight support on ground reaction forces and metabolic power during running. J. Appl. Biomech. 24, 288–297. Hoffman, M.D., Donaghe, H.E., 2011. Physiological responses to body weight–supported treadmill exercise in healthy adults. Arch. Phys. Med. Rehabil. 92, 960–966. Kline, J.R., Raab, S., Coast, J.R., Bounds, R., McNeill, D., de Heer, H.D., 2015. A conversion table for running on Lower Body Positive Pressure (LBPP) treadmills. J. Strength Cond. Res. 29, 863–868. Kluitenberg, B., Bredeweg, S.W., Zijlstra, S., Zijlstra, W., Buist, I., 2012. Comparison of vertical ground reaction forces during overground and treadmill running. A validation study. BMC Musculoskelet. Disord. 13, 235. McNeill, D.K., de Heer, H.D., Bounds, R.G., Coast, J.R., 2015. Accuracy of unloading with the Anti-Gravitys Treadmill. J. Strength Cond. Res. (in press) Millslagle, D., Levy, M., Matack, N., 2006. Kinematic assessment of treadmill running using different body-weight support harnesses. Percept. Mot. Skills 103, 607–618. Newman, D.J., Alexander, H.L., Webbon, B.W., 1994. Energetics and mechanics for partial gravity locomotion. Aviat. Space Environ. Med. 65, 815–823. Raffalt, P.C., Hovgaard-Hansen, L., Jensen, B.R., 2013. Running on a lower-body positive pressure treadmill: VO2max, respiratory response, and vertical ground reaction force. Res. Q. Exerc. Sport 84, 213–222. Ruckstuhl, H., Kho, J., Weed, M., Wilkinson, M.W., Hargens, A.R., 2009. Comparing two devices of suspended treadmill walking by varying body unloading and froude number. Gait Posture 30, 446–451. Sainton, P., Nicol, C., Cabri, J., Barthelemy-Montfort, J., Berton, E., Chavet, P., 2015. Influence of short-term unweighing and reloading on running kinetics and muscle activity. Eur. J. Appl. Physiol. (in press) Samaan, C.D., Rainbow, M.J., Davis, I.S., 2014. Reduction in ground reaction force variables with instructed barefoot running. J. Sport Health Sci. 3, 143–151. Warden, S.J., Davis, I.S., Fredericson, M., 2014. Management and prevention of bone stress injuries in long-distance runners. J. Orthop. Sports Phys. Ther. 44, 749–765.
